Realization of algebras with the help of star-products

Realization of algebras with the help of

★-products

Claudia Jambor 1 , Andreas Sykora 2

o :

O . LM U Miinchen, Sektion Physik

(N ■ Theresienstr. 37, 80333 Miinchen

Max-Planck-Institut fiir Physik,
Fohringer Ring 6, 80805 Miinchen, Germany

OO ' Abstract

We present a closed formula for a family of *-products by replacing
y—( . the partial derivatives in the Moyal-Weyl formula with commuting vector

^ ■ fields. We show how to reproduce algebra relations on commutative spaces

OO ' with these *-products and give some physically interesting examples of

, that procedure.

(N

in
o

o

Oh!
>

X

1 jambor@theorie.physik.uni-muenchen.de

2 andreas@theorie.physik.uni-muenchen.de

1

1 Introduction

The general idea of a ★-product is to introduce an invertible map fl from a
function space to an algebra and then to pull back the algebra product between
the two operators 0(/), il(g). This results in the so called ^-product between
the two functions on the function space

f* M9 = n- 1 (n(m( g )) (1)

which is both linear and associative.

The most commonly used ^-product is the Moyal-Weyl ^-product, for which
we take E for the manifold M and parametrise it by TV coordinates x l . Further
let Q % 3 — const. (i, j = l-- ■ N) be a constant and antisymmetric matrix, then

°° (jh\ n

/ ♦ '/ Y. ' • • • ^ <>■■■■ °- f d h ■ ■ ■ 9 jn 9 (2)

is a ^-product between the functions / and g on R*.

To define more general ^-products we take M to be an arbitrary (sufficiently
smooth) finite dimensional manifold. A ^-product on M is an associative, C-
linear product on the space of functions (with values in C) on M given by

f*g = fg + \Bi (/, g) + (±) 2 B 2 (f, g) + ■ ■ ■ (3)

where / an g are two such functions and the Bi are bi-differential operators on
M . The parameter h is called deformation parameter. There is a natural gauge
group acting on ^-products consisting of C-linear transformations on the space
of functions

f^f + hD 1 (f) + h 2 D 2 (/) + ••• (4)

where the Di are differential operators. They may be interpreted as a general-
izations of coordinate transformations. If D is such a linear transformation it
maps a ^-product to a new one by

f*>g = D- 1 (D(f)*D(g)). (5)

★-products related in such a way are called eqivalent.

If one expands equation (01 in h one gets that a linear transformation D
acting on * only affects the symmetric part of B\

B[(f,g) = B 1 (f,g) + D 1 (fg) - fD^g) - D 1 (f)g. (6)

Since one can show that the symmetric part of Bx may be cancelled by a linear
transformation we may assume B\ to be antisymmetric. Since * is associative,
the commutator

\fU]=f*9-9*f = hB 1 {f,g) + ..- (7)

2

is a derivation with a Leibniz rule

[f*g*h]=f+\g*h] + [f*h]*g. (8)

Up to first order this means that the antisymmetric part of B\ is a derivation
with respect to both functions / and g. Additionally, since the ^-product is
associative, the Jakobi-identity is fulfilled

[/ * [g t h}} + [h*[f* g}} + [g*[h1 /]] = 0, (9)

which implies up to second order that g) is a Poisson structure {. , .} with

{/, {9, h}} + {h, {/, g}} + {g, {h, /}} = 0. (10)

After a certain linear transformation we therefore can always write for the *-
product on a local patch of the manifold

f*g = fg + jrf i a i fd j g + >-- (li)

with

Tl a diTl jk + n fci 5 ; n y + W l diU kt = 0. (12)

Therefore we can classify ^-products up to second order by Poisson structures
on the manifold. The other way around it is even more interesting: if there is
a manifold with a given Poisson structure {.,.}, it is possible to construct *~
products with

f*9 = fg + j{f,g} + ~-- (is)

This was first done for symplectic manifolds (manifolds with invertible II* 5 ') in
■ In pi] a general construction for arbitrary Poisson manifolds is given
(see also pp).

In the following we construct an associative (proof in section ^-product
which provides a generalization of the Moyal-Weyl product J5J by replacing the
partial derivatives with commuting vector fields, since they have similar prop-
erties. For this we give some guiding examples in two dimensions to see that
even in this simple cases interesting algebra structures arise. Furthermore we
propose a method to calculate ^-products that reproduce - even for more com-
plicated deformed algebras - the algebra relations in the according commutative
spaces. With this method we get closed explicit formulas for the ^-product for
a generalization of M(so a (n)), which might be useful for theories like [HH] and
also for the physical relevant M(so q (3)).

3

2 Algebras and ^-products

The space of functions on M. N together with the ★-product Q forms a represen-
tation of the algebra

A = C<x\---,x N > /{[x\x>] -ih0 ij ), (14)

since [x l * a;- 7 '] = ihQ %3 . It is interesting to find ^-products for other relation-
defined algebras like algebras with Lie algebra structures

[x\x 3 ] = ihC t3 k x k , h, C ij k e C (15)

and quantum space structures pi EH with

x*& = qR lJ k ix k x\ q = e h ,R ij kl G C. (16)

Instead of considerering these special relations we discuss a more general
case. We assume that the algebra A is generated by N elements x % and relations

[x i ,x j ]='c\x) = ihc ij (x) (17)

where we assume that the right hand side of this formula is containing a parame-
ter h and that it is getting small as this parameter vanishes. The mathematically
correct context would be a /i-adic expanded algebra

A = C<x\ ■ ■ ■ , x N > [[h]] /{{x\ x j ] - ihc l3 {x)) (18)

where it is possible to work with formal power series in h. Note that this kind
of algebras all fulfill the Poincare-Birkhoff-Witt property since a re-ordering
of two x l never affects the polynomials of same order in h. An algebra with
Poicare-Birkhoff-Witt property possesses also a basis of lexicographically or-
dered monomials. For an algebra generated by two elements x and y this means
that the monomials x n y m constitute a basis.

3 ^-products with commuting vector fields

Let X be a vector field. It is easy to show that

B

• (19)

X\x)^-(J{x) 9 {x)) = { X>(y)^ + X*(z)^-)(f(y) 9 (z))

y — >x,z — >x

To write the rhs in a more compact way we introduce the following notation

Xxhgi = (X 2 + X 3 )f 2 g 3 \ 2 . (20)

With this we can derive a Leibniz rule

X{f m = (X 2 + X 3 ) l f 2 g 3 \ 2 ^ 1 ^ 1 (21)
P(X 1 )f 1 g 1 = P(X 2 + X 3 )f 2 g 3 \ 2 ^ 13 ^ 1 (22)

4

where P is a polynomial in the vector fields. The last equation can also be
written in the form

P{Xi) (/:

253

2->l,3->l

P(X 2 +X 3 )f 293 \ 2 ^ 1 ^ 1

(23)

We now take n commuting vector fields X a = X l a di, i.e.[X a , Xb] — 0. In this case
locally a coordinate system y a (x) with X a = d y a always can be found. Globally
this does not have to be the case. Further let a ab be a constant matrix. Then
we define a ^-product via

X ^f 2 g 3

2->l,3->l

(f*g)\i ■= e

To prove the associativity we calculate

= e

(24)

4->3,5->3

& a '' X a2 (X bi +X b5 ) j^ e <T cd X c4 ,X d5 ^

2-»l,3->l

4->3,5->3.2^1,3^1

(25)

/0 & ab X a iXb2+& a X a iXh3 „CT C X c 2Xdc

2->l,3->l

and

(U*9)*h)\ x

,<T ab X al X b2 I a" d X c3 X,

7354

3->l,4->l

3 (T a (X a 3+X a 4)Xb2 p G C Xc^XdA

gCT afa X a iX b 3-\-<T ab X a 2X b3 ^(7 Cd X c \X d -

3— >1,4— >1,2— >1

/l 52/13

(26)

2-»l,3->l

where in the second step we used the relation \2'.\\\ . The two expressions are
equal since the vector fields commute according to our assumption, so our new
★-product is associative.

For an antisymmetric matrix a we get for the ^-commutator

[fig] = (e° abx ^-e-° abx ^)A 92

= 2sinh( ( T ab X al X b2 )/ 1 . g2 | 2 _ 1 . (27)

In the case of two vector fields, which we call X± = X and X 2 = Y, the explicit
formula for a 12 = h, a 21 = 0, the asymmetric ^-product, reads as

h

/*5 = E^( X "/)(^<7),

n=0

while for a 12 = \,o 21

2 we get

2 n nl

n=0 i=o

(28)

(29)

which yields the antisymmetric ^-product. Both ^-products have the same Pois-
son tensor IP-? = a ah X l a A Y b j .

5

4 Linear transformations

If we have already a ^-product, we have seen that we can produce a new *~
product just by a linear transformation D on the space of functions Suppose
that D is such an invertible operator and that its expansion in derivatives starts
with 1. We here assume, that D is of the form

D = e< x ^ , IT 1 = e- T W (30)

where r is a polynomial in the vector fields X a . The X a are still the commuting
vector fields we used in the last section. Then for the ★-product <24ll we see that

f*'g = D~ 1 (D(f) * D(g))

= e -r(X al ) ( e ^X a2 X b3e r(X a2 ) he r(X a3 ) g3
- p -T(X a 2+X a ; i )+a ab X a2 X b : i +T(X a2 )+T(X a , z )

(31)

V2.93

2— 1,3—1

For t only quadratic in the X a (note that t 2 b can be assumed to be symmetric,
since according to the assumption the vector fields commute)

r = T?X a + \rfX a X h (32)

we have

r(X al ) + r{X a2 ) - r(X al + X a2 ) = -rfX al X b2 (33)
and the new ^-product becomes

f*'g= e^ a f l92

(34)

We see that the antisymmetric ^-product J 2 Oil and the asymmetric ^-product
l|28|i are related by a linear transformation in function space and therefore are
equivalent.

5 Examples in two dimensions

We choose two commuting vector fields and calculate the relations of the it-
product algebra with the asymmetric ^-product

X = xd x ,Y = d y

We take h = ia and get with that

X*X = X 2 , y-ky^y 2 , (35)

x*y = xy + iax, y *x = xy,

6

which yields

[x * y] — iax. (36)

This is the algebra of two dimensional a-euclidean space 0] Q2] . This algebra is
not isomorphic to the two dimensional Heisenberg algebra [x, y] = ih.

X = (a + bx)d x , Y — (c + dy)d y

For the general linear case we do analogous calculations and get

[ x *y]=e bd (y + ~)*(x + ^), (37)

which is an interesting quantum space like structure. This example includes the
case of the Manin plane.

-■ xd y - yd x

of the coordinate transformation into spher-
ay/x * x + y * y. (38)

X = ^= 2 {xd x + yd y ) ,Y =

These are the derivatives d r and dg
ical coordinates r and 0. We get

[x*y] =

X = a(xd x + yd y ) , Y = xd y - yd x

This is a simplification of the previous case. We get

[x * y] = (tana)(x-kx + y*y), (39)

This algebra does not have the Poincare-Birkhoff-Witt property if we would set
tana = 1. Therefore a has to be treated as a formal parameter.

6 Reconstruction of algebras

According to @ the ★-commutator of a ^-product is a Poisson tensor up to first
order

[fl9} = h{f, g} + 0(h 2 ) = g) + 0(h 2 ), (40)

where II is the Possion-bivector of the Poisson structure. This we can get from
the relation

{x i ,aP} = ^ i =c ij (x), (41)

On the other hand, assume that this ★-commutator reproduces the algebra re-
lations, then the rhs of the previous equation would be a ^-polynomial in the
generators of the algebra, i. e.

[x i *x j ]=hci j {x), (42)

7

To calculate the leading order of c±(x) it is not necessary to know the explicit
form of the ^-product, since it always starts with the ordinary product of func-
tions, so we can substitute c+(x) by c(x), which is the same as in ((111) .
Now for the special case of the ^-products l(23jl the Poisson structure is

n = a ab X a A X b . (43)

If we are able to write a general the Poisson structure I14H in this form, we can
reconstruct the algebra relations with the help of the ^-product Il24t . We now
need a general method to find commutative vector fields with the properties
postulated in the previous sections. For this let / be a function and Xf = {/, ■}
the Hamiltonian vector field associated to /. Then the commutator of two such
vector fields is

[X f ,X g ]=X {f<g} , (44)
due to the Jakobi identity of the Poisson bracket. If we can find functions with

{fi, 9j} = Sij, {fi, fj} = 0, {g t , gj} = 0, (45)

this implies that all commutators between the associated Hamiltonian vector
fields vanish. This functions need not to be unique. One can deduce from the
splitting theorem for Poisson manifolds QJ] that this is always possible in a
neighborhood of a point if the rank of the Poisson tensor is constant around
this point. If we do not want to find a ^-product on R , but a ^-product with
certain commutation relations, we can reduce M. N by the set of points where the
rank of the Poisson tensor jumps and we have a good chance to find functions
with the desired properties on the new manifold. In this case we can write the
Poisson tensor as

R = Yl X f* AX 3i- (46)

i

In the following we present functions fi and gt for Poisson tensors of several
algebras and use the corresponding Hamiltonian vector fields in the ^-products
We calculate the resulting algebra relations [. * .] and compare them to
the original algebra relations.

7 Examples for reconstruction of algebras

The two examples we present here are just the most telling ones. The other
ones like for the q-deformed Heisenberg algebra|TJ| , the Lie algebra so(3), the
quantum spaces M(so q (l, 3))[12 , M{so q {A)) ©EU and M(so q (l, 3)) 011121 can
be found in (El 121-

7.1 The quantum space M(so a (n))

Here we investigate the quantum space introduced in J^j. Although this quan-
tum space is covariant under the quantum group SO a (n) 7 we never use this

8

property. We took it because of its simple relations. Further it has a nontrivial
center. Since we are using the n-dimensional generalisation introduced in
we will simply call it SO a (n) covariant quantum space or abreviated M{so a {n)).
The relations of this quantum space are

[x°,x l ]=iax l for i ^ 0, (47)

with a a real number. The x l simply commute with each other. In the following
of the example Greek indices run from to n — 1, whereas Latin indices run
from 1 to n — 1. This algebra is a higher dimensional generalization of the two
dimensional algebra we calculated in section We take as the manifold
and use coordinates x° and x l . With the two dimensional case in mind we guess
the following vector fields

X = iax^i, Y = d Q . (48)
For the asymmetric ^-product l(28jl we get with this vector fields

[x* * x } = iax\ (49)

Even more useful is the generalization of the above algebra with the new
relations

[x a ,x p ] =i(a a x f3 -a x a ), (50)

where a a are now n different deformation parameters. For this relations to be
consistent the Jacobi identities have to be fulfilled, which easily can be proven.
Since the rhs of the relation is linear and we are therefore dealing with a Lie
algebra the Poisson tensor associated with the algebra is just

{x a ,x f> } = a a x f> -aPx a . (51)

If we want to find commuting vector fields that reproduce this Poisson tensor we
now follow the way outlined in sectional The rank of this matrix is 2. Therefore
we have to find two functions f,g fulfilling {f,g} = 1. We make a guess and
define

f = a a x a i a = x a - ^-x 13 (52)

with a 2 = a a a a . These functions fulfill commutation relations very similar to
the special case of M{so a (n)):

{/, Z a } = a 2 x a {Z a ,x p } = 0. (53)

If we define g = -\ln V x a x a we find

{/,<?} = !• (54)

9

The commuting vector fields are now

X = {f,-} = a 2 x^d p - (a a x a )a' 3 5 /3 (55)
Y = {-,g} = -a' 2 a dp. (56)

In this case no singularities have shown up and the ★-product can be defined on
whole R™. Again we may use the asymmetric ^-product (l28l) and see that the
algebra relations are reproduced.

7.2 The quantum space M(so q (3))

Our second example is the ^-product for the quantum space M(so q (3)), which
is invariant under the quantum group SO q (3) |12| . The algebra relations in the
basis adjusted to the quantum group terminology are

zx + — q 2 x + z, zx~=q~ 2 x~z, [x~ , x + ] = (q — (j _1 )z 2 . (57)

and therefore the Poisson brackets are

{z,x + } = 2zx + , {z,x~} = -2zx~, {x',x + } = 2z 2 . (58)

For / = ilnir~and g = Inz we get the desired {/, g} — 1, so the Hamiltonian
vector fields become

z 2

X f = zd z + — d+, X g = 2{x+d+-x-d-). (59)
For the ^-product we take

cvz 2

X = zd z + — d+, Y = x+d+-x-d^. (60)

For the asymmetric ^-product II28|I we get the following algebra relations

z*x + = e h x + *z, (61)
z-kx" = e~ h x~ -k z, (62)

[x + *x~] = 7j(e~ 2h - l)z*z. (63)

If we set

2 1

ea, a= - — T (64)

q + q- 1

this reproduces the original algebra relations.

10

For the antisymmetric ^-product Il29fl the algebra relations become

z*x + = e h x + -kz, (65)
z-kx" = e~ h X~*z, (66)
[x + *x~] = ~{e h -e- h )z*z. (67)

Again we get back the original algebra relations, if we set

h 2

o = e2 a = -. 68

q + q- 1

Acknowledgements

We want to thank Julius Wess for the possibility to work on this topic and his
steady support. Also we want to thank the MPI and the LMU for their support.

References

[1] Alberto S. Cattaneo and Giovanni Felder. A path integral approach to
the kontsevich quantization formula. Commun. Math. Phys., 212:591-611,
2000.

[2] B. L. Cerchiai, R. Hinterding, J. Madore, and J. Wess. A calculus based
on a q-deformed heisenberg algebra. Eur. Phys. J., C8:547-558, 1999.

[3] M. DeWilde and P. B. A. Lecomte. Existence of star-products and of formal
deformations of the poisson lie algebra of arbitrary symplectic manifolds.
Lett. Math. Phys., 7:487-496, 1983.

[4] Marija Dimitrijevic et al. Deformed field theory on kappa-spacetime. Eur.
Phys. J., 031:129-138, 2003.

[5] Marija Dimitrijevic, Frank Meyer, Lutz Moller, and Julius Wess. Gauge
theories on the kappa-minkowski spacetime. 2003.

[6] L. D. Faddeev, N. Yu. Reshetikhin, and L. A. Takhtajan. Quantization of
lie groups and lie algebras. Leningrad Math. J., 1:193-225, 1990.

[7] B. V. Fedosov. A simple geometrical construction of deformation quanti-
zation. J. Diff. Geom., 40:213-238, 1994.

[8] Claudia Jambor. Dissertation. 2004.

[9] Achim Kempf. Quantum group symmetric bargmann-fock construction.
1992.

[10] Achim Kempf. Quantum group symmetric fock spaces with bargmann-fock
representation. Lett. Math. Phys., 26:1-12, 1992.

11

[11] Maxim Kontsevich. Deformation quantization of poisson manifolds, i. 1997.

[12] A. Lorek, W. Weich, and J. Wess. Non-commutative euclidean and
minkowski structures. Z. Phys., C76:375-386, 1997.

[13] S. Majid and H. Ruegg. Bicrossproduct structure of kappa poincare group
and noncommutative geometry. Phys. Lett., B334:348-354, 1994.

[14] H. Ocampo. so q (4) quantum mechanics. Z. Phys. C, 70:525, 1996.

[15] W. B. Schmidke, J. Wess, and B. Zumino. A q deformed lorentz algebra.
Z. Phys., C52:471-476, 1991.

[16] Andreas Sykora. The application of star-products to noncommutative ge-
ometry and gauge theory. 2004.

[17] Alan Weinstein. The local structure of possion manifolds. J. Diff. Geom.,
18:523-557, 1983.

[18] Julius Wess. g-deformed heisenberg algebras. 1999.

[19] Julius Wess and Bruno Zumino. Covariant differential calculus on the quan-
tum hyperplane. Nucl. Phys. Proc. Suppl., 18B:302-312, 1991.

12